luisarojas / student-matching Goto Github PK

View Code? Open in Web Editor NEW

9.0 3.0 1.0 14.61 MB

A web application designed to match mentors and students based on a series of survey, personality-related, questions.

License: MIT License

Python 14.02% HTML 5.50% CSS 72.92% JavaScript 6.80% Makefile 0.33% Shell 0.20% Dockerfile 0.23%

webapp personality matching mentors mentees students matching-algorithm survey questionnaire neo4j

student-matching's Introduction

Background && Motivation • Overview • Algorithm • Features • User Interface

Background && Motivation

Background

Ontario Tech Student Mentorship Program

Motivation

Overview

✅ Automated group generation per faculty through the matching of mentors and mentees.
👉🏼 Excel file upload and download.
👉🏼 Ability to merge batches of students entered.
👉🏼 Custom question weighting.
👉🏼 Manual mentor and mentee assignation.

✅ Mentor log management and mentee feedback for program assessment.
👉🏼 Form builder and response manager.

✅ Track the continued engagement of mentees and mentors.
👉🏼 Download of tracked engagement data summary.
👉🏼 Visualization of engagement per group.

Tools && Architecture

Algorithm

We looked into different current algorithms that could be a good fit for our problem. However, we found significant drawbacks for those that we found, so we opted for designing our own:

✅ Based on K-Means clustering algorithm
✅ Customizable number of mentees per group
✅ Guarantees one mentor per group

Features

Matching

Data Management

Utility

Analytics

User Interface

student-matching's People

Contributors

Stargazers

Watchers

Forkers

thriledlokki983

student-matching's Issues

For question weights input, replace the current text boxes with sliders or dropdowns.

The options for the user to pick from should be the following: "Low", "Medium", "High".

Fix bug where an error is outputted if the upload is clicked again after the initial submission

Check that the type of answers for the same questions in both mentor and student files are also the same.

Check that the mentor and student raw files both contain the exact same Faculty names.

Update multicoloured bar to display real faculty distribution metrics

Resize dropdown element for faculty filtering

The graph database should not return an error if there are no matches currently stored in the database.

Instead, just return nothing, or a message stating there is nothing to display.

Hide dropdown content when the user clicks outside of it

Check that we only take into account questions that are BOTH, in the mentors and the mentees files.

Last Match - violation causes table to wipe clean

Cannot figure out how to replicate this bug:
[Violation] 'setTimeout' handler took 82ms

Have environmental variables in a separate config file

This will be important specially once we get started with the databases.

Calculate Euclidean distance with the exponential value of each vector element instead.

Make all success codes in requests same

Grab HTML input field names dynamically.

Instead of hardcoding them in the server, they should be passed with the rest of the data in the request.

Remove grey row highlight when all rows are selected in the Last Match table.

Refactor the functions from match.py and clean-data.py that are being called from the server.

Implement filtering by role (mentors/mentees)

By default, select the first row in the Last Match table

Highlight mentors in the resulting output file.

Fix resizing bug in the Faculty column of the Last Match table

Fix Glyphicons in the sidemenu

The Glyphicons being used right now seem to belong to an older version of Bootstrap, so they're not being shown. We should update them to work with version 4.0.0.

Change "Edit Matches" button in Step 2 to "View Matches"

When manually assigning mentees, permanently highlight their current mentor in the dropdown

Modify categorical to numerical transformation in order to have them as ordered.

Currently, pandas assigns numerical values to categorical data in an unordered fashion. Since we're giving specific weights to question answers, we need them to be in order (e.g. strongly disagree - strongly agree would correspond to 0-4 respectively).